文章摘要

文章信息/Info

Title:: Design of noise control algorithm for CT image of pulmonary tumor based on Bayesian rough set

Author(s):: ZHUO Deqiang; Department of Radiology, Zhongnan Hospital of Wuhan University, Wuhan 430071, China

Keywords:: Bayesian rough set; Lung; Reduction; Tumor CT image; Wavelet coefficient; Anti-noise

摘要:: 贝叶斯粗糙集处理噪声数据能力强，分类肺部肿瘤CT图像结果准确，为图像去噪提供精准的图像分类结果。基于此，设计基于贝叶斯粗糙集的肺部肿瘤CT图像抗噪算法，基于贝叶斯粗糙集分类模型进行肺部CT图像分类，约简贝叶斯粗糙集属性和决策规则，基于决策规则预测肺部CT图像类别；对存在肿瘤的CT图像噪声小波系数构建拉普拉斯数学模型，基于贝叶斯最大后验概率估计小波系数概率密度，计算噪声方差和子代小波系数标准差，使去噪算法具备自适应性；基于小波系数的概率密度得到最大后验（maximum a posteriori,MAP)估计值，对该值做小波反变换，实现肺部肿瘤CT图像自适应去噪。结果表明，该算法去除肺部肿瘤CT图像噪声效果好，抗噪能力强，较好保留图像细节特征，视觉效果佳。

Abstract:: Bayesian rough set has a strong ability to process noise data and classify the CT image of lung tumor accurately, providing accurate image classification results for image denoising. Therefore, the anti-noise algorithm of lung tumor CT image based on Bayesian rough set was designed. The lung CT image classification was conducted based on the Bayesian rough set classification model. The attributes and decision rules of the Bayesian rough set were reduced. Laplace’s mathematical model was established for the noise wavelet coefficients of CT images with tumors. The probability density of the wavelet coefficients was estimated based on the Bayesian maximum posterior probability, and the noise variance and standard difference of the wavelet coefficients were calculated to make the denoising algorithm adaptive. The maximum a posteriori(MAP) estimation value was obtained based on the probability density of the wavelet coefficient, and the value was converted into the inverse wavelet transform to realize the adaptive denoising of the CT image of lung tumor. Experimental results show that this algorithm has good noise removal effect, strong anti-noise ability, better retention of image details and visual effect.